The internet of things (IoT) is beginning to happen. More items are being connected to the internet daily and this will have a major impact on the datacentre.
IoT is a scenario in which objects, animals or people are able to transfer data over a network and talk to each other without requiring human-to-human or human-to-computer interaction.
Gartner estimates that the IoT will have 26 billion units installed by 2020 and, by that time, IoT product and service suppliers will generate incremental revenue exceeding $300bn, mostly in services.
"IoT deployments will generate large quantities of data that need to be processed and analysed in real time," says Fabrizio Biscotti, research director at Gartner.
"Processing large quantities of IoT data in real time will increase the workloads of datacentres, leaving providers facing new security, capacity and analytics challenges," he adds.
The majority of datacentre traffic is self-generated through applications hosted on servers in the facility, with a degree of traffic between facilities, whether these are co-location or public-cloud services.
But, as machine-to-machine (M2M) communication – technology enabling networked devices to exchange information and perform actions without the manual assistance of humans – increases, datacentres will transform.
Already, data levels caused through access by users from standard devices (PCs and laptops) is being outweighed by the data from mobile devices (tablets and smartphones), often accelerated through a bring-your-own-device (BYOD) grassroots move.
This move towards a more mobile workforce has caused many issues in the datacentre already. These include increasing vectors for security attacks, through public networks and via additional native apps on devices, as well as a greater amount of “chatty” data between the various devices and the datacentre.
What IoT means to the datacentre
More on the internet of things
- The internet of things is set to change security priorities
- APIs key to security of internet of things, says Axway
- Internet of things to power classroom education
- Smart Grid and the internet of things
- Smart technologies and the internet of things
- Gartner: Internet of things will be worth trillions
- Explained: What is the internet of things?
- Building internet of things applications with DeviceHive
- The internet of things – the devices are taking over
- ARM buys Sensinode for internet of things push
Yet, imagine what it will be like for the datacentre as the IoT rush begins.
The internet of things could introduce a slew of new device types to work in the enterprise.
More of the production line will be directly connected to the corporate network, rather than being a separate, proprietary network, as seen in many places. Security systems will become more digitised – CCTV, employee tracking and so on, will all move to being used across the standard TCP/IP Ethernet network.
Intelligent buildings, where temperature and other environmental aspects are monitored at a granular level, on a regular basis – to ensure a constant environment is maintained for workers – will also be plugged in to the network.
The datacentre will also soon have to deal with GPS systems monitoring an organisation’s vehicle fleet, as well as radio frequency identification (RFID) tags monitoring corporate assets, among many other devices.
And it won’t stop there. Already, the line between work and home is blurred. The digital home will provide individuals with the means of controlling their own life – but from a mix of their own and corporate devices.
If an organisation uses any form of centralisation for employee IT use, such as virtual desktops, this data could also be flowing through the datacentre.
Datacentre bandwidth – the first IoT victim
The immediate impact is on the bandwidth to and from the datacentre. The IoT will mostly be external to the datacentre when it comes to data creation and, on the whole, the data created will be in small chunks. But the IoT will be, by its very nature, extremely chatty.
Imagine 1,000 devices creating data on a per-10-second basis – and imagine how that might affect your network bandwidth and, therefore, the overall performance of applications as perceived by the employees and other users.
Now, think what it could be like if there were 10,000, 100,000 or a million devices – these numbers are well within the reach of the growth of the IoT, as it moves towards the internet of everything (IoE).
Is your datacentre ready to face data traffic of this magnitude?
The first thing enterprises need to consider is that, just because something can be attached to the internet, it doesn’t mean it should be.
The canteen drinks machine could be connected to an organisation's network – but, as it is probably already connected to the distribution company’s network via a telephone line and lets them know when it needs to be refilled, there is no need for the organisation's network to bother about it.
A business should ensure it has a proper business reason as to why something should be part of its network of the IoT.
Another tip is to look at the IoT as a hierarchical system. Not all data from every device needs to come to the datacentre.
By taking a more local-intelligence approach, the data traffic to the datacentre can be minimised. For example, with an intelligent building, does every single reading of temperature, humidity, infrared and so on, need to be sent through?
With the use of an intermediary controller, the noise level of standard readings can be ignored, with only the outliers of possible problems getting through to the datacentre.
Securing the IoT
To secure datacentres to accomodate an enterprise's IoT initiatives, IT must:
- Employ multi-layer pattern matching to identify threats as far out towards the devices as possible.
- Not let the threats get too close to the datacentre.
- Develop a new approach to security as low-power, low-maintenance and low-cost periphery devices that can provide an adequate first line of defence will be required.
The same applies to security systems. If an employee’s security badge and other credentials match their position, the datacentre doesn’t need to know. Only if there is a problem with matching everything up should a more powerful central capability need to be brought in.
Choosing how often different data needs to be monitored is also important. Making sure the time period is correct for the need can cut down the noise level drastically and make the IoT easier to manage.
Enterprises could also look to offload certain data. For example, employee data coming from their own IoE networks should not have to go through the datacentre.
Enforcing sandboxing on their devices – whether this is a PC, a tablet or a phone – creates a secure environment for corporate data and systems, which is separate to the environment where the user runs their own apps.
Sandboxing is an IT approach to software development and mobile application management (MAM), which limits the environments in which certain code is executed.
Datacentre managers should force data they can identify as being personal to traverse a separate network that only talks to the personal partition of the device.
This should preferably be one the individual is paying for, such as a mobile connection. However, where this cannot be enforced, a virtual slice of the corporate network that bypasses the datacentre should be used, maintaining bandwidth for where it is needed.
Priority and quality-of-service measures should be applied to all data streams, ensuring that applications that are mission-critical and latency-sensitive can throttle down, or even stop IoT data streams, if required.
To prepare datacentres to face the IoT, IT teams should also look closely and identify the many new end points, each of which is an attack vector for new security threats.
The IoT is marching up on organisations and they need to be prepared, or they may see performance suffer at every level.
This was first published in August 2014